Re: bosh update - monit - related - query


Danny Berger
 

Make sure your pre-start script will exit with an error if a command fails
(like `set -e`). I would guess your pre-start script is calling `monit
start {job-name}`, but that will error since the jobs haven't yet been
installed. Instead, call your `/var/vcap/jobs/{job-name}/bin/ctl` script
directly and avoid monit.

I ran into the issue once as well and raised a GitHub issue at
https://github.com/cloudfoundry/bosh/issues/1094 before learning the
alternative.

On Wed, May 18, 2016 at 3:55 AM, Nishtha Srivastava <nishwish.0311(a)gmail.com
wrote:
Hi Dmitriy,

The pre-start script seems to work fine during bosh deployment update
(pre-start script gets called before job's ctl script).
However during first deployment, pre-start script does not run. And the
deployment fails.

Any pointers here will be appreciated.


Regards,
Nishtha


On Mon, May 16, 2016 at 2:05 PM, Dmitriy Kalinin <dkalinin(a)pivotal.io>
wrote:

http://bosh.io/docs/pre-start.html might be what you need.

Sent from my iPhone

On May 16, 2016, at 1:33 AM, Nishtha Srivastava <
nishwish.0311(a)gmail.com> wrote:

Hi all,

I have created a bosh deployment in which during processing of job,
there is one step which involves copying of data.
Time taken for this step is dependent on the size of the data. When the
data size is huge, copying takes several minutes and job updation fails.
<job_name> is not running after update

The only way forward then is to unmonitor the process on the deployed
vm, run the deployment again and start monitoring.
I know that Monit will wait up to 30 seconds for the start/stop action
to finish before giving up and report an error.
I guess this is the problem in our case.

Seek your expertise here.
What could be the way forward for such a long running job in bosh
release.



Thanks & Regards,
Nishtha
--
Danny Berger

Join cf-bosh@lists.cloudfoundry.org to automatically receive all group messages.